Diffusion models can be used to lossily compress images for real, see for example Lossy Compression with Gaussian Diffusion (https://arxiv.org/abs/2206.08889). I would expect this to work with text-conditional diffusion models too and give you better compression when supplying a correct caption.
Dang it! I knew this existed but I couldn't find it for some reason. Thank you!!! This has been a wonderful journey...my poor kids have been getting spammed with my latest incantations for the past week lol.
I keep trying to think of ways to make this more consumable for the less technically inclined, but GPUs are just not cheap. I was on a V100 for probably 36 hours through Colab via my $9 monthly subscription, that'd run $100 on AWS for the same amount of time on a p3. So then I'd need to figure out a tip jar or something to keep the gas tank full.
After AlphaZero, DeepMind went on to develop MuZero (https://en.wikipedia.org/wiki/MuZero, paper: https://arxiv.org/abs/1911.08265), which is able to learn to play games without being provided an explicit model of their rules. "When evaluated on Go, chess and shogi, without any knowledge of the game rules, MuZero matched the superhuman performance of the AlphaZero algorithm that was supplied with the game rules."
Discord has already had it for a while. Some users seem to appreciate the mandatory cooling-off period while others feel it makes it harder for them to respond when a bunch of users pile on them in quick succession.
> because the parser generator knows about all patterns you want to match in advance, it will match longer terminals before shorter—more ambiguous—terminals
If you don't mind manually ordering your choices in your alt operators, then PEG and parser combinators are OK.
Oh, and you often need to restructure your grammar to avoid left recursion.
"...perhaps we may have to compromise with sin and provide a hard copy terminal after all."
giggles
"In the second place, you will see that the new information system will make the public more responsive to the careful reasoning of you good guys and more immune to the blatant propaganda of those bad guys."
Up with Good! Down with Bad!
(But seriously, futurism generally isn't this uncannily accurate. I know hindsight lets me select the best predictions a posteriori - anyone know any way of improving this? - but it's still interesting.)
